Skip main navigation

Hurry, only 11 days left to get one year of Unlimited learning for £249.99 £174.99. New subscribers only. T&Cs apply

Find out more

Discrete choice experiments

An introduction to discrete choice experiments and how and why it is used in AMS.
The word choices written on a piece of card, with arrows emerging in many different directions.

This activity was developed by Dr Liz Morrell, a senior researcher at the University of Oxford, and provides an introduction to discrete choice experiments and discusses how they can be applied in AMS projects to help design interventions and policies.

To begin, we will introduce the discrete choice experiment (DCE) and the type of research question it is used to answer. We will proceed to discuss the steps for developing and executing a DCE, and examples of DCEs used in recent work related to antimicrobial resistance.

Introducing the discrete choice experiment (DCE)

A DCE is a quantitative, survey-based technique that is used to understand how individuals value the various features of a product or service and their strengths of preference. It involves asking respondents to state their preferred choice between a number of presented alternatives. Those alternatives are described in terms of the features – known as attributes – of the product or service. From the choices respondents make, we can determine which of those attributes influence the choice, the relative importance of the attributes, and the trade-offs being made between them.

DCE’s are described as a stated preference technique – that is, one where you are asking participants to state a hypothetical choice. This contrasts with a revealed preference study, where you would observe the actual choices made by people in real life.

Results from DCEs can be used, for example, to identify the characteristics that will be important for the success of an intervention, and the required level of these characteristics. As a stated preference technique, DCEs are very important in healthcare, which is not a free market and has limited opportunities to use revealed preference studies. The technique is also useful in understanding the potential for innovative ideas, where the product or service does not yet exist and revealed preference work is not possible; the DCE allows us to present hypothetical alternatives.

Example 1

A screen-reader compatible PDF is available here.

Here’s an example of a typical discrete choice question. At the top, we provide respondents with the context in which we’re asking them to make a decision. They are then presented with two or more alternatives, described in terms of five features, and asked to choose which one sounds best.

The example illustrates the terminology used in describing DCE’s. The features listed on the left are known as attributes, and each of them can take a defined number of different levels. Each of these profiles is referred to as an alternative, and the alternatives along with the decision we ask them to make, is a choice question.

Notice that respondents are being asked to make trade-offs between these alternatives or competing priorities, for example, consultation B is longer, but patients have to wait longer to get the appointment. These allow us to estimate their relative importance. Notice also that there are both continuous and categorical attributes, for example, time is continuous and which GP you will see is categorical.

Example 2

A screen reader compatible version of this table is available here.

In this example, the study is exploring preferences for approaches to lockdowns during the Covid pandemic. The table on the right shows the range of levels used for each of the attributes. Respondents often find concepts like probability difficult to understand, so complementing them with simple graphics is common.

The question being asked here is: what do you think your state governor should choose. The study acknowledges that the respondent isn’t the decision-maker, but asks for their preference for a community that they are part of. This perspective that we’re asking the respondent to take can affect how they respond, so it is an important consideration in designing the experiment. We can distinguish three types of perspective:

  • Personal: for themselves (as in the first example)
  • Social: for others
  • Socially inclusive: for themselves and others (as in this example)

Example 3

If you require a screen-reader compatible PDF, this is available here.

This final example examines preferences among the public and patients with rheumatoid arthritis, for a prescribing aid described as a ‘biological calculator’. The aim is to learn how each of the features contribute to the attractiveness of the calculator, and the circumstances in which they would be attractive.

Importantly, the researchers offer a status quo option – that is, not using a biological calculator at all. The levels for each attribute in the status quo option – included if the DCE is intended to predict uptake of a product or intervention – are set to reflect what we do today and will be constant across all choice questions.

In the comments section below, please let us know:

  • What is your level of familiarity with discrete choice experiments?
  • Describe a research question in your work environment that could be addressed using a DCE.

Click ‘next’ whenever you are ready to learn about how DCEs are used in practice.

This article is from the free online

Tackling Antimicrobial Resistance: A Social Science Approach

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now